Discourse Theory

Bohman James. Handbook of Political Theory. Editor: Gerald F Gaus & Chandran Kukathas. 2004. Sage Publication.

‘Discourse theory’ may be used in a broad and in a narrow sense. In the broad sense it refers to any theoretical enterprise that considers language in use, that is, language as it is used in practices and performances, from the analysis of ordinary conversation or public argumentation to formal scientific papers or parliamentary discussion. It can also be used in the narrow sense of a particular sort of normative ethical and political theory derived from the work of Jürgen Habermas (1984; 1996). This theory provides an account of those social practices in which dialogue, reason giving and argumentation play a central role and which thus may be called ‘discursive.’ It is also a theory of rationality based on the practical know-how of speaking and acting subjects that is a social scientific alternative to instrumental or strategic conceptions dominant in rational choice and game theory. Discourse theory in both senses has already found wide application from argumentation theory (Crosswhite, 1996) to the sociology of scientific knowledge (Lynch, 1993). It has also become significant in political theory, especially in constructive approaches to the public sphere and democracy on the one hand (Habermas, 1996; Calhoun, 1989) and in critical analyses of race and gender on the other (Goldberg, 1990; Butler, 1993). While the former figures prominently in normative theories, the latter sort of discourse theory is often the basis for showing the inadequacies of normative claims to reason and justification.

Discourse in the broad sense includes ‘talk,’ ‘writing,’ and ‘discussion.’ Social scientists use discourse theory to analyse how people talk about politics and social problems (Gamson, 1992). Discourse theory may also attempt to uncover shared assumptions and assumed capabilities, such as the ways in which specific policy issues (such as nuclear power or international trade) are framed in terms of the extent of expert authority. Such assumptions can shift to a more publicly oriented frame (Gamson, 1988). Political discourses also emerge around institutions, as when constitutions and constitutional courts produce an evolving discourse on the nature of rights and obligations in liberal democracies (Dryzek, 2000), or Orientalism around European colonialism (Said, 1978), or when various international financial institutions give loans and create policies to promote ‘development’ based on specific models of ‘baskets’ of human goods as commodities (Sen, 1999). These discourses themselves can become the subject of second-order public debate and discussion, as critics and citizens become dissatisfied with such policies and the assumptions that guide them. In this case, discourse becomes a means not merely for conveying information or for public discussion, but rather for the contestation and challenge of policies and practices.

How can discourse do all these things? When applied to politics, discourse theory focuses on practices that have features that go beyond mere talk. Discourse in political practices and in the public sphere seems to be directed to an implied audience or ‘unseen gallery’ and thus goes beyond ‘sociable’ interaction among friends (Gamson, 1992: 20). Thus, discourse is communication directed to an indefinite audience, and an extension of face-to-face interaction that is made possible by technologies of writing, mass media or computer assisted communication and by formal political institutions (Thompson, 1995). Second, discourse that has the property of being public is also reflexive or second-order communication; it must at least include the possibility of communication about the mode and assumptions of communication itself, for example, whether it is really public or not (Habermas, 1984). This reflexivity is apparent especially when communication fails, when the assumptions that we make for practical purposes ‘until further notice’ in Garfinkel’s (1969: 33) phrase are no longer successful in producing mutual understanding or co-ordination of action. In this case, speakers must make explicit the basis of communication itself by providing reasons and arguments that others might be able to accept. Just how far the demand for justification can be pursued by speakers and institutionalized in practices is subject to dispute among the proponents of various theories of discourse. For some, the linguistic medium makes reflexivity possible, while for others it imposes insuperable limits on reflection (Hoy and McCarthy, 1994).

For these reasons, discourse theory emerges at the intersection between philosophy, social science and political theory. On the one hand, various disciplines in philosophy underwent the ‘linguistic turn,’ especially in the philosophy of language where the slogan ‘meaning as use’ replaced less socially informed theories. This emphasis on use focuses attention on ‘how we do things with words’ and thus also on the conditions of success for various sorts of speech acts (Austin, 1962; Searle, 1969; Habermas, 1979). At the same time, the philosophy of social science attempted an ‘interpretive turn,’ in which agents’s own self-interpretations become central to making actions intelligible to us, rather than explaining actions in light of causal laws or mechanisms available to the observer. In interpretive social science, texts and discourse become central objects for interpretation (Geertz, 1973; Taylor, 1985), while the focus on shared meaning led to the rejection of the emphasis on individual preferences and aggregation common to rational choice theorists. In social theory, then, these turns gave new importance to the potential generative role of communication and the structuring role of language and linguistic practices.

Political theory has experienced a similar set of turns. The role of discourse highlights the differences between the ‘market’ and the ‘forum,’ between the aggregation of given preferences in social choice mechanisms and the formation and transformation of preferences in public discourse (Elster, 1997; Cohen, 1997). Nowhere was the shift to discourse more important than in democratic theory. This distinction allows political theory to take a ‘deliberative turn’ in emphasizing discursive and communicative practices in which participants attempt to convince each other by offering reasons in public discussion and debate. Not only does this open up a space for giving and asking for reasons in the public sphere or in various forums, but reason giving is also a particularly non-coercive form of political integration and a potentially effective method for solving problems and settling conflicts. All of these various ‘turns’ in philosophy, social science and political theory together make discourse central to normative and empirical theorizing and mark a watershed in thinking about the form of democratic politics and social integration specific to modern societies.

The focus of this chapter is on the various uses of discourse theory that are now common in political theory. First, several approaches to discourse theory need to be distinguished along three important dimensions: whether they are normative, empirical, or both. Next, the usefulness of discourse theory will be illustrated with respect to normative democratic theory, with respect to institutional design and to democratic deliberation. Third, the critical uses of discourse theory will be developed in terms of problems of ideology and toleration that are not merely limited to ‘non-ideal theory’ in Rawls’s (1999) sense. Finally, I consider the limits of discourse theory and suggest that the issue is properly epistemic rather than linguistic, a matter of the precise nature of the critical know-how necessary to participate in discursive practices. The proper goal of such a discursive political theory is to avoid the impasses of past debates: the Scylla of an empty idealization of discourse and the Charybdis of a blind scepticism that offers no guide to the practices in which discourse is employed.

Approaches to Discourse Theory: Normative and Empirical

Discourse theory has been developed through three competing approaches. The first and broadly ‘constructive’ approach is fundamentally normative, where the practical know-how of speaking and acting subjects is developed into a theory of communicative rationality that has implications for how we ought to think of political and legal institutions (Habermas, 1984; Rawls, 1999). It construes discourse as a rule-governed activity, the rules of which may be reconstructed as procedural idealizations (such as giving all the opportunity to speak, to engage in all forms of speech and so on). Such a theory permits political theorists to develop explicit rules for governing discourses, rules that may have either a role in criticizing existing discursive practices or a constructive role in evaluating and designing institutions. Since not all assumptions of discourse can be made fully explicit in rules, such an account can no more be a complete account of democratic political life than a written constitution can describe all of its derivative practices. If such a theory is too idealizing, there may be a wide gap between the norms and ideals it proscribes and the existing practices. Faced with this gap, other theories of discourse try to capture deeper, more structural assumptions and presuppositions that shape actual discussion and practices (Foucault, 1977; Bourdieu, 1991; Butler, 1993). This approach identifies deep linguistic structures and thus eschews explicit rules, aiming instead to uncover deep practical constraints operating through norms. It alerts us to relations of power within discourses.

These two conflicting approaches are not the only available theoretical options. The third, broadly ‘reconstructive’ and critical approach combines the best features of both (Bohman, 1996; Hoy and McCarthy, 1994). It seeks a theory that is normative without relying solely on idealizations and counter factual ideals, and empirical without becoming sceptical of all attempts to institutionalize discursive practices of justification. Such an approach is operative in some proponents of the deliberative turn in democratic theory. A defensible discourse theory thus provides a test case for normative theory that is informed by social science but still seeks to develop robust and practical norms for guiding institutions and practices.

Constructing Ideal Discursive Procedures

As the leading proponent of the normative theory of discourse, Habermas proposes that the development of norms of discourse is the task of a theory of communicative or discursive rationality, where rationality is defined as ‘how speaking and acting subjects acquire and use knowledge’ (1984: 11). Such a broad definition suggests that the theory could be developed through explicating the conditions for reaching understanding through language, and this task falls primarily on ‘formal pragmatics.’ ‘Formal pragmatics’ is Habermas’ term for a general account of the capacity of a speaker to use and understand speech acts correctly: ‘the know-how of subjects who are capable of speech and action, who are attributed the capacity to produce valid utterances, and who consider themselves capable of distinguishing (at least intuitively) between valid and invalid expressions’ (1990: 31). The focus of formal pragmatics is on the know-how necessary for producing and evaluating correct and incorrect expressions or valid and invalid utterances, or for producing well-formed utterances that meet the conditions of successful communication.

What might such a formal pragmatic analysis contribute to a theory of discourse? The intuitive knowledge of a competent speaker permits them to engage in second-order evaluation in asking for justification or reasons for various sorts of validity claims that are implicit in utterances; to understand an utterance is to know its ‘acceptability conditions.’ While validity claims may remain implicit so long as communication is unproblematic and ongoing, competent speakers may also demand that the implied warrant be redeemed and demand explicit justification in second-order communication (communication about communication, or ‘discourse’ proper) in order to reach an understanding. Habermas locates the rational potential of communication in discourse in the explicit and second-order capacities of actors to provide reasons for their own claims and evaluate the reasons offered by others; they thereby engage in argumentation, through which the implicit basis of ongoing communication is suspended and made the basis of explicit testing, judgement and assent. Such second-order communication is discourse when it takes the form of acts of communication that suspend the constraints of action and co-ordination and examine the validity claims implicit in the utterances made by speakers.

Such a reconstruction of implicit know-how may have a critical function in so far as it can specify when speakers violate the conditions of rationality implicit in communicatively successful utterances. For Habermas, reconstruction also has a constructive role to the extent that these conditions can be explicated and then formulated as explicit rules or principles. A formal pragmatic theory could then reconstruct discursive justification in a general way, through what Habermas calls ‘the principle of discourse.’ When applied to normative statements, this principle offers a proceduralist justification of any norm in the form of a principle of universalization: ‘Only those norms of actions are valid to which all those affected could agree as participants in rational discourses’ (Habermas, 1996: 138; Baynes, 1995: 208). The general principle of discourse is then specified in a principle for the justification of norms or rules. Second-level principles of discursive justification can then be applied in various more specific principles tailored to specific domains of discourses, such as moral or legal argumentation or the variety of forms of political deliberation (Alexy, 1989). These explicit principles guide practice, and institutions in various domains ought to approximate them in justifying their rules or actions.

Habermas’s explicit rule for democratic legitimacy is analogous to Kant’s ‘general principle of right,’ in that the principle of democracy is a general principle of legitimate law making: ‘Only those laws are legitimate that can meet with the agreement of all legal consociates in a discursive process of law making that in turn has been legally constituted’ (1996: 141). The democratic principle is then an application of discursive justification applied to the law making process. Laws are valid as norms to the extent that those subject to law also formulate and agree to them as participants in rational discourse. In this discursive process, citizens are the authors of the laws to which they are subjected; they are legally guaranteed certain rights that ensure public and private autonomy. Discourse theory not only permits us to recast general normative principles in terms of discursive procedures, thus enriching Kant’s principle of right or Rawls’s first principle of justice as equal freedom; it also has a constructive role in formulating principles that guide or regulate the very practices it reconstructs. The principle is both ideal and proceduralist: the conditions of legitimacy are counterfactual. Under such ideal conditions of assertibility, all participants in the discourse would not only agree, but would agree for the same reason, so that all disagreement must be due to the ways in which the actual conditions fall short of the ideally rational procedure. This abstraction from actual discourse leads some to propose a historically contingent and context-specific theory rather than a theory of discursive rationality.

Empirical and Sometimes Sceptical Approaches

The second set of approaches start from a less idealized and more empirical view of discourse, finding in it a constraining and limiting rather than an enabling condition for reflection and deliberation as essential aspects of social practices. Social scientific approaches look closely at the specific features of discursive contexts, noting for example the role of social status in the emergence of the scientific discourse (Shapin, 1994) or the role that slavery and racial categories play in the discourse on citizenship in American history (Smith, 1997). Here we find that the closer we look at actual discourses, the more they depart from the ideal procedural conditions that constitute their rationality. Moreover, ethno-methodological discourse analysis tied to specific situations shows that norms and rules are highly flexible and contingent in their application in informal contexts, even if they are crucial to the act of making others intelligible (Heritage, 1984). The critical legal studies movement has shown that many legal and constitutional norms are indeterminate, and even capable of justifying decisions that now seem to contradict them (Unger, 1986). Indeed, empirical studies demonstrate how rules shape practices (when they do) and are useful in closing the gap between ideal counterfactual analysis and its application to social and political practices, as well as in showing why procedures may fail to realize the discursive principles on which they are purportedly based (Hoy and McCarthy, 1994). However, some empirical analyses of discourse attempt to underwrite more sceptical challenges to normative theories. There are two main critics who take these empirically informed challenges a step further in language related to social power: Michel Foucault and his theory of ‘discursive formations’ as regimes of truth, and Pierre Bourdieu and his conception of symbolic power.

Michel Foucault argues discourses can be analysed as ‘regimes of rationality’ which are not independent of power and its effects but rather are constitutive of a ‘general politics of truth.’ It is on the basis of such a regime rather than ideal discursive conditions that speakers accept that something is a ‘truth candidate’ (Hacking, 1986). In this way, validity is not independent of social context but is relative to a regime of truth that shapes what is possibly true, normatively correct or practically feasible. Moreover, the human sciences are themselves inscribed in a regime of truth that is also implicated in social technologies that establish the normal and the abnormal, the distribution of bodies in social space. Foucault (1977) argues that the effects of power discourses are connected to ‘disciplinary practices’ that cannot be dissolved by democracy. Here the issue is one of agency: whether or not discourse is something so deeply constitutive that it is no longer under the control of speakers.

By contrast, Bourdieu’s challenge is more epistemic, relativizing linguistic activities and practices to a background habitus, a set of dispositions inculcated in socialization. The object here is to appeal to ‘generative and implicit schemata’ rather than explicit or consciously sanctioned rules. Practices are regular and reproducible patterns of action ‘without being the product of rules and without presupposing a conscious aim or the express mastery of them’ (Bourdieu, 1977: 55). He criticizes ideal theories for their ‘linguistic communism,’ as blind to the forms of status and inequalities that make it possible for speakers to be authoritative and persuasive. The capacity to produce comprehensible utterances ‘may be quite inadequate to produce sentences that are likely to be listened to, likely to be recognized as acceptable in all situations in which there is occasion to speak’ (Bourdieu, 1991: 55). Normative discourse theories leave out social relations among speakers, their different social positions and their capacities to garner linguistic authority. Bourdieu thinks that because habitus is not a matter of rules, its limitations are not in principle accessible to speakers at the level of second-order communication, when speakers must offer explicit justification for their actions and practices. Both challenges see power as operating within discourse itself, not merely as an external constraint upon it. These same sorts of constraints on discourse may also operate in the ways that deeply historically embedded inequalities such as race and gender shape discourse and restrict its reflexivity (Butler, 1993: 232).

These challenges to normative theories of discourse raise important questions about the epistemic constraints on speakers and social limitations on the linguistic medium. This sort of limitation may in part be overcome by the formal organization of speech in institutional settings, such as in courts of law or democratic institutions of parliamentary debate. They also must be answered at the same level at which they are raised: the analysis of the restrictions on communication and more importantly on discourse as the second-order communication in which justification of practices and policies occur. Normative theories of discourse discuss these same issues in terms of the theory of ideology and the critical attitude of toleration in communication. Before turning to these problems as limitations on democratic practices, let me turn first to the discursive reconstruction of democratic theory, including questions of institutional design.

Discourse and Democracy

Any discursive account of democracy is not merely an account of democratic discourse, however idealized it may be. Rather, it must itself be a complete account of democracy, in the sense of offering a reconstruction of its usual elements while giving them novel interpretations. In what follows, I will discuss the main lines of a reconstruction of democracy as a discursive practice, guided to a large extent (although not exclusively) by Habermas’ normative political theory. Although Habermas, Dryzek and some others use the terms ‘discursive’ or ‘communicative’ democracy or offer a ‘discursive theory of democratic legitimacy,’ almost all theories of deliberative democracy have to a large degree been shaped by discourse theory proper or offer an implied theory of discourse themselves. This discursive component defines what deliberation is to be; for example, ‘reasoned argumentation,’ or discussion guided only by ‘the force of the better argument,’ where decision making must be based on reasons that ‘all may accept as free and equal citizens.’ Indeed, John Dewey already argued that democracy itself is not a feasible idea unless there exists ‘full publicity,’ or free and open communication necessary for deliberation as a form of social inquiry. Whatever obstructs or restricts publicity, he argued, ‘limits and distorts public opinion and checks and distorts thinking on social affairs’ (Dewey, 1988: 339). How might institutions approximate this ideal and promote full publicity?

Discursive Designs

Discourse theory has a properly constructive role in providing the basis for various forms of institutions designed with the aim of creating opportunities for wide and effective participation in discursive processes of public discussion, deliberation and argumentation. Constitutions are in part discursively designed, so as to establish not only the separation of powers and thus discursive competence, but also a division of labour in communication and deliberation. Broadly speaking, the framers of the United States Constitution had deliberation in mind, in designing institutions that would produce the ‘mild voice of reason’ that would overcome narrow self-interestedness, the passions, and the mischief of factions (Bessette, 1994). More recent discussions of the deliberative or discursive design of democratic institutions reflect a three-level distinction of various aspects of political life in a complex and pluralistic modern society. Such a society is differentiated in a number of ways, with distinctions between the state and the market, civil society and its associations, and the political public sphere of citizens and various sub-public spheres. In general, discourse and deliberation can go on both within and outside various formal institutions, in civil society and the public sphere as well as in the formal institutions of the modern state with its law making powers and authority (Habermas, 1996; Dryzek, 1996).

The discursive approach to democracy leads to an institutional design that is based on a ‘two-track model,’ in which on the one hand formal institutions generate effective decisions through the medium of law and thus are ‘jurisgenerative’ (Michelman, 1988; Habermas, 1996), and on the other the robust public sphere and civil society allow citizens to engage in deliberation with each other from a variety of perspectives. Deliberative politics then takes place in both tracks at once, in a complex discursive network that includes argumentation, discussion, bargaining and compromise. Formal institutions must be designed to be open to influence from the wider and more informal public sphere and civil society, with various mechanisms such as representation and elections that ensure not only access to influence but also that a variety of perspectives emerge in deliberation and debate. Formal institutions require at least the widespread perception of legitimacy, and in this way ‘cannot operate without an associated and supportive discourse (or discourses)’ (Dryzek, 1996: 204). This includes the discourse of rights and citizenship for liberal institutions or the discourse of scientific authority and expertise for many administrative institutions.

A main issue separating various proponents of deliberative designs is whether or not and in what ways public deliberation actually shapes or should influence decisions. Such influence may be direct or indirect. Others want a more direct role for deliberation, seeing institutional reform of law or administration as necessary in order to make them more open to citizens’ deliberation; this would require new forms of decision making, including deliberative planning or citizen juries. Thus, there is a conflict between a view of public discourse as providing challenges to formal legal and political authority and, as such, being indirectly deliberative (Dryzek, 2000; Pettit, 1998) and subject to discursive challenge from the outside; and the view of those who see it as more directly deliberative in the decision making process itself (Habermas, 1996: ch. 8; Dorf and Sabel, 1998). This is not as much of a forced choice as some make it out to be, since in some instances challenge may be the best or indeed the only effective means for influence given the way in which political authority is constituted; or there may be cases in which more directly deliberative approaches are necessary to preserve the reality of popular sovereignty and accountability to citizens. Scientific or expert authority that is delegated public power provides an example of the first; the planning process in public administration provides an example of the second. Indeed, there seems to be a continuum from direct to indirect deliberation, depending on the sort of institutions and supportive discourses involved.

Democracy and Administration: Designing Non-Democratic Institutions Discursively

The discursive design of democratic institutions seeks to open the policies and decisions of powerful institutions to discursive testing. Newer forms of political authority such as expertise and the media seem to operate outside the potentially discursively designed constitutional state and are less open to discursive influence. Administrative institutions act for the common good, a use of public power authorized by legislative mandates to achieve certain ends. For that reason, philosophers from Locke to Hegel and Weber see administrators as engaged only in ‘neutral’ means/ends reasoning, a necessity for the exercise of effective political power. Foucault and others have analysed the way in which this power is exercised in part via discursive means, in the way that people and things are named, classified and disciplined in a ‘symbolic order’ (Foucault, 1977; Bourdieu, 1991; Flyvbjerg, 1998). Social scientists also have long recognized the ambiguous relationship between democracy and bureaucracy: Weber saw that democracy helps produce more bureaucracy, even as bureaucracy tends to undermine democracy as the former becomes an efficient ‘social machine’ (Weber, 1946; Hummel, 1994), open only indirectly to deliberative influence.

The alternative is to put deliberative mechanisms and interaction with the public within the design of administrative institutions themselves, and this sort of design has taken the form of ‘deliberative planning’ (Fischer and Forester, 1994; Forester, 1993). As Habermas puts it, administrators ‘cannot avoid appealing to normative reasons when implementing legal imperatives,’ so these processes must occur within procedures that pass the test of constitutional legitimacy and lead to the ‘democratization of administration’ (1996: 440) by discursive means. Similar sorts of considerations might apply to other forms of non-democratic social authority, such as the authority of medical researchers that has been recently challenged by AIDS activists who sought to directly influence the practice of medical experimentation (Epstein, 1996).

These collaborative processes could certainly be fruitfully applied to deliberative processes within other institutions of the constitutional state, making them all potentially more ‘directly deliberative’ than the two-track solution to size and complexity permits (Dorf and Sabel, 1998). Discursive modes of decision making are more feasible if decision making power is dispersed, where implementation is not subject to the requirement of uniform solutions and thus open to local variations and concerns. In fact, such decentralized and directly deliberative processes seem appropriate in supranational contexts that go beyond the representative institutions of the modern state, such as the emerging post-sovereign polity of the European Union.

Democratic Discourse: Restricted or Plural in Form?

For some proponents of deliberative democracy, a strong distinction between reasoned argumentation and mere discussion provides the basis for the claim that deliberation must be oriented to consensus (Habermas, 1996; Cohen, 1997). Deliberation is not merely discourse or dialogue, Cohen argues, because it must be ‘reasoned,’ that is based on ‘public argument and reasoning among equal citizens’ that yield the single best answer (1997: 74). Critics often charge that both of these claims are exclusionary and lead to undemocratic consequences under the circumstance of background injustice and pervasive inequalities. It might seem that an orientation to consensus is not a requirement of deliberation, even if it may function as a regulative ideal. Deliberation must at least resemble argumentation to the extent that it is a matter of giving and asking for reasons. The reasons that make a decision acceptable ought to be distinguished from modes by which they are communicated. Democratic standards demanded for decisions need not apply to the medium of communication as such, and not all formal public spheres need to be ideally inclusive. This means that formal theories of communication and rationality cannot decide in advance precisely what modes and forms of communication are empirically appropriate in various settings.

The first tension between the empirical and normative dimensions of a theory of democratic deliberation concerns whether or not there is a specific type of discourse that characterizes democratic deliberation in general, as Habermas and Cohen hold for argumentation. Once again, it appears that the choice is between a theory of deliberative democracy that takes deliberation to be highly constrained and thus potentially exclusionary, and one that takes deliberation to have no normative constraints in the informal public sphere. The rejection of the former may be motivated by the attempt to see public deliberation as broader than the confines of formal institutions of the constitutional state (Dryzek, 2000). Any specific form of discourse may privilege certain citizens over others, as when argumentation favours articulate and dispassionate speakers and thus the better-educated elites (Young, 2000; Sanders, 1997). It seems an empirical question whether argument favours the privileged. Regardless of how this debate about acceptable forms of discourse is settled, even more important for political equality is the fact that some differences in competence and abilities among participants will remain. Assuming that both formal and informal settings are necessary for robust deliberation, how could formal and informal discourses interact so that unjust privilege and unequal influence may be avoided?

Is argumentation really a formal mode of discourse? If discourse is to be distinguished from acts of communication as a second-order and reflective activity, then argumentation in a general sense is the mode of critical self-reflection, of making claims and justifications explicit. Furthermore, if utterances make validity claims and these claims are supported by reasons, then argumentation is precisely the process by which speakers’ claims can be tested and made explicit (Habermas, 1984: 42). Even here, however, there remains an irreducible empirical diversity of types of argumentation, from the strict arguments made in the context of scientific disciplines or the regulated context of a court with rules of evidence, to arguments that attempt to convince ‘anyone.’ Rather than being merely formal, argumentation can be seen rhetorically as a way to settle conflicts over reasons and assumptions that inform practices, although less than in the conception of persuasion through oratory favoured by some critics of deliberative democracy (Remer, 1999). As opposed to both formal and rhetorical models, such an account conceives of arguments dialogically, as the giving of reasons and the answering of objections raised by one’s fellow citizens. Rather than as a means of reaching a conclusive agreement, argumentation is better seen as an ongoing means of resolving conflict that is successful only if each perspective is taken into account and each objection given a hearing (Crosswhite, 1996: 102ff).

Similar criticisms emerge when reason giving is thought of by critics in an overly cognitivist and consensualist way. With regard to the first, reasoned argumentation is often construed logically as linking premises to a conclusion in a complex series of statements that is not enthymematic. While some reason giving may be guided by institutionalized, strict requirements such as in a court of law, reasons are better construed as discursive responses to challenges to claims: ‘A claim is not an argument; a claim with a reason is’ (Crosswhite, 1996: 79). Reason giving and argumentation may be seen not only in a more dialogical way, but also as operating in the specific context of disagreement and conflict and their resolution. Argumentation makes the conflict explicit and mutual, establishing an exchange of challenges and reasons between the claimant and respondent (1996: 102ff). On this view, there are special features of all ‘public’ reasons; if all participants may raise challenges, this responsiveness must be oriented to an indefinite audience and is still possible even given persistent disagreement. Indeed, disagreement is precisely what makes democratic deliberation not only necessary, but also fruitful and productive when tested through the variety of perspectives typical of a diverse and pluralistic audience. Argumentative discourse need not presuppose unanimity, or seek consensus, but rather places conflicts within a mutually constructed space of reasons.

This fact of disagreement raises the issue of whether or not public deliberation is ‘oriented to consensus.’ Consensus is meant here to contrast with mere aggregation of preferences in voting and with bargaining or compromise. Certainly, if democracy were only voting and bargaining, it would lack the self-critical testing and responsiveness of reason giving and discourse; the problems of the tyranny of the majority and aggregation problems of social choice would undermine the effectiveness of democracy and its claims to legitimacy. However, if we demand too much agreement and an overly strong conception of consensus, then we lose the advantages of resolving conflict through argumentative means. Habermas thinks that participants in argumentation must be guided by the ideal of a single right answer to which all agree ‘for the same reasons’ (1996: ch. 8; Bohman and Rehg, 1996). He may well be correct that an overly agonistic conception of public discourse would undermine the epistemic basis for claims to democratic legitimacy, that is, that democratic deliberation is legitimate and not only is a fair process, but is more likely to find the most equitable and true outcome (Estlund, 1997). For all its attractions to critics of deliberation, agonistic debate is no less open to the charge of elitism (Benhabib, 1991), and even less based on the sort of co-operation needed to resolve conflict mutually. At the same time, the demand that all agree for the same reasons is overly strong and reduces the epistemic benefits of argumentation and challenge. If participants agree for different reasons, the epistemic gains that result from testing any agreement from a plurality of perspectives would arguably be superior to the gains of any orientation to consensus as a regulative norm.

Besides issues related to the emphasis on argumentation and consensus as overly narrow, other critics of deliberative democracy argue that it has too narrow a conception of the range of discursive possibilities within the public forum, leading to the exclusion of rhetoric, testimony, and other important modes of speech that do not seem to be forms of reason giving. The wider notion of argumentation as involving claims, challenges and reasons as responses vitiates these criticisms to some degree. If ‘all speech acts must be open to all participants’ in free and open communication, then perhaps the most important deliberative speech act is related to the opening of a discursive exchange or the proposing of a topic or theme for public deliberation (Bohman, 1996). Indeed, to make a claim is to invite a response, and with this kind of invitation comes an implicit obligation to be responsive to those who reply. Indeed, the discursive obligations of citizenship involve not only the willingness to engage in the special mutual conflict distinctive of argumentative practices but also obligations of responsiveness and answerability to others. Listening is thus just as important an obligation as speaking, and it is here that asymmetries are likely to emerge rather than on the expressive side, however formally restrictive some public spheres may be in permissible modes of expression. What if such collaborative perspective taking is blocked, and communication remains unsuccessful in resolving conflict? This raises issues of ideology and toleration, of putting the current and sometimes unnoticed limits of discourse up for democratic debate and challenge.

Ideology and Democracy: Toleration and the Limits of Discourse

Democracy traditionally refers to a specific set of institutions that assure citizens’ self-rule via procedural mechanisms that, at the very least, permit equal access to political influence. For example, making decisions according to voting rules such as the formal principle of ‘one person, one vote’ is also an attempt to assure political equality by distributing political power widely. Other decision rules would require different forms of equality: in ‘deliberative politics’ in the constitutional state, equal chances to participate in deliberation might be conjoined with mechanisms of decision making by majority rule (Habermas, 1996: ch. 8). However important they may be, the necessary conditions for deliberative politics are not exhausted by explicit rules of justification or the distribution of power in decision making. Besides the background of common knowledge of such rules and of a shared political culture, democracy in general and deliberative democracy in particular require a particular communicative infrastructure. Without the effective operation of implicit norms of communicative success as a resource available to all, formal procedures and institutions, no matter how well designed, will not succeed in distributing power in accordance with explicit norms of political freedom, equality and publicity (Bohman, 1996: ch. 3). The lack of consideration of the relation between implicit norms of communicative success and explicit norms for the distribution of power has led to practical deficits in normative theories of discursive politics. Here we can incorporate the insights of sceptical empirical theories and apply them to the existing structures of communication and deliberation in particular institutions. If they are to be made grounds for deepening democratic practices, these should be formulated not as theoretical claims about the limits of language or reflection in general, but as the efficacy of citizens who currently lack effective voice and address their criticisms to other citizens as claims to justice. Is democratic discourse a means of overcoming the implicit restrictions of political discourses informed by social categories of racist and sexist speech?

Rather than only being a set of explicit principles of justification and institutional decision rules, democracy is also a particular structure of communication. It is a structure of communication among free and equal citizens. By contrast, ideology restricts or limits social processes of communication and the conditions of success within them. As a reconstruction of the correct insights of the Marxian critique of liberal ideology, the theory of distorted communication is therefore especially suited to the ways in which meanings are used to reproduce power even under explicit rules of equality and freedom. This is not to say that explicit rules are unimportant: they make it possible for overt forms of coercion and power to be constrained, the illegitimacy of which requires no appeal to norms implicit in practices. For example, violations of communicative freedom may remain implicit: the success of a deliberation may simply not be a matter of putting one’s reasons up for evaluation by others when one avoids communication altogether. Under conditions of great inequality, contested topics may simply be avoided at the agenda setting stage that reflects organizational bias.

In any actual democracy, both strategic and communicative action may be present. For example, large advantages in the agency freedom of one group over all others may be due to the possession of vastly greater resources or other forms of social power; the achievement of their goals may not depend upon the consensual resolution of a conflict with groups with less social power. If Przeworski and Wallerstein (1988) are right, for example, powerful economic groups have historically been able to attain their agency goals not by explicitly excluding topics from democratic discussion but rather by implied threats and other non-deliberative means (Bohman, 1996). We can see the differences between such strategic forms of interaction to the extent that they reflect differences in bargaining power, regardless of the democratic means used to reach this equilibrium. Threats of declining investments block redistributive schemes, such as those that would burden well-off groups with higher tax rates; these credible threats circumvent the need to convince others of the reasons for such policies or to put some issue under democratic control. Similar discursive effects occur when institutions operate with implicit discursive frames, as did the Nuclear Regulatory Agency when it considered the 1966 partial meltdown of the Detroit Edison reactor to be a mere ‘engineering mishap’ (Gamson, 1992). The excessive agency freedom of some and the lack of social power of others means that some dissenting reasons will not become topics to be recognized or respected. However, it is possible to shift the framework of justification in both these cases, where the meanings of policies are changed and new agendas formed. In these cases, strategic actions by social movements are used to open up communication where it is blocked, to move discourse and deliberation beyond a bargaining equilibrium asymmetrical negotiating power.

By looking at such cases, we can better see the division of labour in reconstructive theories of discourse. Explicit rules function to create the frameworks in which institutions operate to the extent that they can be embodied in deliberative procedures. But this constructive role for the theory is not sufficient, since implicit social norms can undermine communicative success within an institutional framework of explicit rules. Civil rights, for example, may be interpreted legally so as to establish and guarantee a minimum threshold and the fair value of communicative liberties. They can be interpreted, for example, to assure that voting power is more equitably distributed, permitting greater access to representative forums, or they may open up regulations of political speech to diminish the effects of discrepancies in campaign financing. The emergence of new norms or the reinterpretation of old ones may require a period of what Ackerman (1991) calls ‘constitutional politics’ within an existing democracy. Ackerman thus sees the constitution as an open-ended discursive project subject to paradigm shifts at historical junctures such as Reconstruction after the Civil War and the Great Depression. These changes reflect ‘discourse moments,’ to use Gamson’s (1992: Part I) term, in which the people, the courts, or the executive respond to historical circumstance by reinterpreting and recreating the Constitution.

Besides constitutional reform, limitations on expression may demand the formation of alternative public spheres, the developed forms of expression of which expand the pool of reasons and the styles of acceptable public communication in the larger public sphere. In all of these cases, the critic is equipped with the reflective abilities of a participant in a communicative process, not the least of which entails the ability to challenge the correctness of the communicative process itself. But in this case, circularity is avoided because the critic does not have to start from scratch: bootstrapping of new communicative possibilities begins with the ability to participate in those areas of everyday communication, no matter how small, which are not distorted by power. At the very least, reflection produces gains in freedom by permitting speakers to become aware of the ways in which implicit violations of norms limit public functioning and inhibit those very corrective and transformative performances that might change the conditions of communication.

This possibility of self-critical communication requires that the virtue of toleration be given a discursive reconstruction and brought to bear on the problem of ideological restrictions on deliberation and communication. In a context of a high degree of social and cultural pluralism and the conflicts that it might engender, proper responsiveness would require toleration among citizens, even if such toleration were extended to taking their perspectives seriously while challenging their claims in public deliberation. In this way, toleration is required in order that we treat others as political equals, as having equal entitlement to contribute to the definition of the society in which they live (Scanlon, 1996). In order to capture the obligations of public deliberation, Onora O’sNeill (1990) correctly argues that it is communication itself that is ‘the proper object of toleration’ in a democracy. In deliberative settings, citizens manifest their equality with each other not only by refraining from interfering with their acts of expression, but also by sustaining the conditions for communication. How do they do this? They do this reflexively, in their communication with each other in public deliberation and in their attitudes towards others as participants in a public process (Bohman, 1999). Toleration in this sense is discursive openness.

If publicity is the more general norm and attitude of concern for the structures and processes of communication in a democracy, then toleration demands that citizens be concerned with the structural features of public debate and discussion through which deliberation takes place. Toleration in a weak sense is directed towards the reasons that others offer in communication: they must be taken seriously and not disqualified ex ante (either in principle or in fact). Toleration is needed in the public process aimed at discovering whether a reason is a publicly acceptable one or not. Publicity in this sense is practical and historical rather than merely a formal ideal. If the public character of a reason in this sense is better seen as an outcome of an actual process of discussion, then it is not necessarily significant if the reason is religious or secular (Rawls, 1999). However, taking reasons seriously is not all that deliberation requires. Toleration in the strong sense extends not directly to reasons as such but to the perspectives that inform these reasons and give them their cogency. Before a reason can first be seen as a reason and then potentially as one that passes the critical scrutiny of all citizens, the perspectives of others and the experiences that inform them must be recognized as legitimate; in light of this inclusion of their perspective, groups recognize themselves as contributing to democratic decisions. The toleration of others’ perspectives is then part of recognizing them as equal members of a political community, despite the potential for persistent disagreements and deep conflicts. As Scanlon (1996) puts it, what toleration expresses is recognition of common membership that is deeper than these conflicts, recognition of others as just as discursively ‘entitled as we are to contribute to the definition of our society.’

These two features of toleration – as perspective taking and as normative attitude in communication take up the sceptical challenge of putative limits of discourse. A regime of toleration is illegitimate if it denies discursive entitlements by falsely generalizing the perspective of the tolerating group so that they can reject the claims and reasons of the tolerated group. A regime of toleration is just if it permits citizens to fulfill their obligations of justification to all if they are to respect the equal entitlement of each to contribute to the definition of their society. The toleration of perspectives is a matter not only of first-order communication, but of the second-order properties of the regime that aims at protecting the integrity of communication and deliberation (Young, 1997; 2000). In this respect, toleration is a second-order property of the framework that creates a deliberative community. It is also a property of citizens, who are obligated to exhibit concern for democratic communication. When coupled with critical reflection on the conditions of successful communication, toleration acts as a form of anti-power to overcome the restrictions of ideology on the structure of communication in democratic processes. Those who make these criticisms may act as the ‘generalized other’ in Mead’s sense, the other whose claims test the limits of the supposedly free and open discursive community of citizens. The limits on discourse are then limits on the regime of toleration and its implied generalized other to whom the regime must be justified.

Conclusion: Extending the Discursive Community

Besides these applications to democratic practices and institutions, discourse has properties that make it a unique medium. It is certainly reflexive and self-referential, since it is primarily through discourse that we can challenge discourses and their implicit restrictions. This reflexivity makes discourse uniquely suited to extending democracy. By permitting indefinitely large and indirect social relations, discourses are not confined to specific linguistic communities but may extend beyond their historical, social and cultural origins. As Mead (1934) put it, ‘the universe of discourse’ is the most inclusive and extensive of all human communities, if that term may be applied to any grouping determined by participation in intensive communicative interaction. The universe of discourse then enables ‘the largest conceivable number of individuals to enter into some social relationship to each other, however indirect and abstract that may be’ (1934: 158). Mead asks further political questions of organizing discourse in institutions: ‘Can we carry on a conversation in international terms? It is largely a question of social organization’ (1934: 271). With global communications, interaction, and media, the global public sphere seems to be a new reality to be reckoned with politically (Bohman, 1999; Habermas, 2000). This provides at least the potential for democratic innovation and for new cosmopolitan discourses, both made possible at least in part by the extension of communicative interaction in the emerging global public sphere.

The emergence of a transnational civil society and a global public sphere provides the countervailing global infrastructure for the same sort of contestation that is the basis for the democratic accountability of the media and technoscience. Cosmopolitan democrats must foster the conditions for communication that make this contestation effective. Their goal has already been formulated in deliberative theories of democracy: to support a communicative infrastructure needed to expand the possibilities of democratic politics to the global arena where asymmetries are now prevalent. Given the scale of such a democracy, influence on decision making may be highly mediated and often indirect. The political structure of this higher-level democracy is yet to be determined, but its demands of scale make it unlikely that it will institutionalize the discursive principle in the same way as it has been in the modern state. For now, cosmopolitan democracy consists mostly of discursive challenges to the current international order from transnational civil society and public spheres, since it is not yet organized institutionally as to permit deliberative authorization or equal opportunities for effective participation rather than contestation. The cosmopolitan ideal is thus another potential target for discursive reconstruction, since in discourse all speakers become the potential addressee of claims made by others to whom we have the rational obligation to be open and responsive. If politics is the means by which a society acts upon itself, then increasing interdependence may lead to the expansion of politics. Since discourse remains the means by which politics becomes deliberative, the question is then to come up with a feasible institutional design to organize discursive exchange in the international political community.